EvoAug-TF: Extending evolution-inspired data augmentations for genomic deep learning to TensorFlow.

Area of Science:

• Genomics
• Computational Biology
• Machine Learning

Background:

• Deep neural networks (DNNs) are powerful tools for predicting molecular functions in the non-coding genome.
• Functional genomics experiments often yield limited data, hindering DNN training.
• Existing augmentation methods like EvoAug are limited to PyTorch, excluding TensorFlow users.

Approach:

• Developed EvoAug-TF, a new package extending EvoAug's evolution-inspired data augmentation to TensorFlow.
• Systematically benchmarked EvoAug-TF against the original EvoAug package.
• Ensured EvoAug-TF is open-source, freely available on GitHub, PyPI, and documented on ReadTheDocs.

Key Points:

• EvoAug-TF enables effective DNN training for genomics using limited functional genomics data.
• The package provides comparable performance to the original EvoAug.
• Extends the utility of EvoAug to the widely-used TensorFlow framework.

Conclusions:

• EvoAug-TF significantly broadens the accessibility of advanced data augmentation techniques for genomic DNNs.
• The tool facilitates improved generalization and interpretability in genomic predictions.
• Open-source availability promotes wider adoption and research in the field.